Upper back stretch

Key points

In one pilot trial, the stretch pushed the straight arms forward and rounded the upper back so the shoulder blades moved apart¹.
In that trial of 50 people with trigger points (tender knots) in the rhomboids (muscles between the shoulder blades and the spine), every group added another technique, myofascial release or isometric relaxation; no group stretched alone¹.
Trials of the extension and rotation variations often tested programmes, such as shoulder-blade and upper-back exercises in 32 office workers². One tested kneeling rotation alone in 24 people with neck pain³.
In a review of healthy people, four stretching studies of the upper back and neck showed no effect on posture; strengthening showed a very large one⁴.
In 24 people referred to a syncope (fainting) clinic with suspected fainting unrelated to stretching, a seated extension variation caused a brief blood pressure drop similar to standing up⁵.

Movement

The named stretch

In one pilot trial (a small first trial that tests whether a larger one would work), the stretch was done sitting or standing¹. The arms reached straight out at shoulder height with the fingers interlocked¹. The arms pushed away while the upper back rounded, so the shoulder blades moved apart, with the chin slightly tucked¹. The trial called it the scapular protraction stretch; protraction is drawing the shoulder blade forward around the rib cage¹.

The same trial used three related upper-back stretches¹:

A seated stretch: arms crossed, hands on the opposite shoulders, leaning forward to round the upper back with the chin tucked¹.
A standing wall stretch: hands on a wall at shoulder height, stepping back and lowering the chest so the shoulder blades spread¹.
A hug stretch: the arms swept forward from the sides to cross at chest level, slowly, three times¹.

The authors say the seated stretch is felt between the shoulder blades and across the upper back¹.

Variations that extend or turn the upper back

Other studies moved the upper back in other ways, mostly bending it backward or turning it. These are variations, not the named stretch^3,6:

Over a roller. In a case report (a report on a single patient), one man bent his upper back backward over a roller, without overextending the neck⁷.
Kneeling rotation. Sitting back on the heels, one hand on the floor and one behind the head, the trunk turned and the elbow lifted³.
A set of office-worker exercises. Rolling the upper back over a foam roller, pushing a Swiss ball forward while sitting, and lifting the upper body lying face down².

The clinic blood-pressure study below used a seated extension form⁵. The shoulders were raised and turned outward while the neck tipped back⁵.

What it loads

The muscles between the shoulder blades

The two rhomboid muscles draw the shoulder blade back toward the spine, raise it and turn its outer border downward⁸. The trapezius, the large muscle across the upper back and neck, has upper, middle and lower groups of fibres^9,10. Its middle fibres draw the shoulder blade in toward the spine⁹. With the lower fibres, they join the rhomboids in drawing the shoulder blades back¹¹. The named stretch moves the shoulder blades apart, the opposite way¹.

The upper back’s joints

In the thoracic spine (the part that carries the ribs), the rib cage and overlapping bony points greatly limit bending forward, backward and sideways¹². A systematic review (a set, checkable search and summary of research) pooled studies of intact cadaver spines, from bodies donated for study¹³. Summed from T1 to T12, the thoracic segments, it found these ranges, smaller than values often quoted from older work¹³:

Range (degrees). Bars start at zero; the table below gives each value with its source.

Movement of the whole thoracic spine	Range (degrees)
Bending forward and back	28¹³
Bending to the side	36¹³
Turning	45¹³

Kyphosis is the outward curve of the upper back. A narrative review (a summary written without a set search method) reports that its angle averages 20 to 29 degrees in a person’s twenties¹⁴. It starts to rise after 40, faster in women than men, reaching a mean of 52 degrees in women aged 76 to 80¹⁴. Another narrative review defines hyperkyphosis as an angle above 40 degrees¹⁴.

The upper-back curve and shoulder pain

Studies of the curve itself and shoulder pain disagree^15,16. A review of 10 studies found moderate evidence of no kyphosis difference between people with and without shoulder pain¹⁵. In a cross-sectional study (one look at a group at one time) of 64 male competitive swimmers, those with shoulder pain had a larger curve¹⁶. The medians were about 23 against 19 degrees¹⁶. The authors say shoulder pain may be associated with a larger curve while raising the arms¹⁶.

In a review of five studies in 624 people, a larger curve went with less shoulder abduction (lifting the arm sideways) and flexion¹⁷. The correlations were about −0.16 and −0.12¹⁷.

Blood pressure in the extension form

A study at a syncope (fainting) clinic tested 24 otherwise healthy people⁵. They had been referred with suspected fainting from causes unrelated to stretching⁵. Each held the seated extension form 10 to 15 seconds⁵. It caused a brief drop in blood pressure similar to standing up, with a smaller rise in heart rate⁵.

On standing, systolic (upper-number) pressure fell by a median 28 mmHg⁵. The heart-rate response to the stretch averaged about 31 percent of that on standing⁵.

Stretch-induced syncope is fainting set off by stretching the upper back, neck and shoulder muscles; a literature review calls it uncommon¹⁸. It found 12 studies describing 30 cases aged 5 to 26¹⁸. In those cases, bending the neck far back and stretching the upper back set off the episodes¹⁸. In a study of 9 people with this kind of fainting and 24 controls, an extension stretch lowered blood pressure in both groups¹⁹. The fall was greater in those with stretch-induced fainting¹⁹.

The authors of the clinic study suggest caution with upper-body stretches, particularly upright ones, for people with orthostatic intolerance (trouble staying upright) or autonomic disorders⁵. The two blood-pressure studies tested an extension form; the named stretch rounds the upper back instead^1,5,19.

How the body adapts

The named stretch, tested only alongside other methods

The pilot trial of the named stretch randomised, by computer, 50 staff of a single institute with rhomboid trigger points¹. Both groups did the stretches, one with myofascial release (hands-on pressure on muscles and the fascia around them) and one with isometric relaxation¹. So the trial compared two add-ons; no group did the stretches alone or had no intervention¹. Isometric means the muscle works but the joint does not move.

Over four weeks	Stretches plus myofascial release	Stretches plus isometric relaxation
Pain score, out of 10	About 7.3 to 5.3¹	About 7.2 to 6.4¹
Arm raise, degrees	About 47 to 52¹	About 47 to 49¹

The authors judged the pain change with isometric relaxation below the threshold that matters clinically¹. No adverse effects were reported in either group¹. The person taking measurements did not know the groups, and the trial had no funding¹. The authors list the small sample and no long-term follow-up as limits¹.

Trials of the variations in neck pain and forward head posture

Forward head posture is a posture with the head held in front of the shoulders. Most trials below tested a programme, not one stretch^2,6:

Who was studied	What was compared	What was found
24 people with long-term neck pain and forward head	Kneeling rotation against neck stabilisation exercise, four weeks	Rotation improved more in pain, range, posture angles and disability³; pain fell 3.75 against 2.50 points³
32 people with forward head posture	Upper-back mobilisation plus a face-down upper-back lift, against neck mobilisation and exercise	The upper-back group did better at six weeks²⁰; the authors judged how much that matters uncertain²⁰
32 office workers with forward head posture	Shoulder-blade and upper-back extension exercises, against neck exercises and stretching, six weeks²	Both improved; pain and one breathing measure differed between groups, head posture angle did not²
26 office workers with long-term neck pain	Upper-back mobility exercises against manipulation, both with neck mobilisation and exercise⁶	Both improved, with no difference in upper-back range, pain or disability⁶

Mobilisation is gentle movement of a joint, often by another person; manipulation is a quick, small hands-on thrust. The office-worker authors could not separate the effects of the shoulder-blade and upper-back exercises². The rotation trial’s authors could not judge long-term effects, because there was no follow-up³.

A meta-analysis pools the results of earlier studies. One review pooled 13 trials in 819 people with forward head posture²¹. Neck programmes and neck-plus-upper-back programmes both improved head posture and disability in the short term²¹. The combined programmes also reduced pain²¹. The certainty of this evidence ranged from low to very low²¹.

Trials in active people and office workers

In 36 teenage female volleyball players, adding four weeks of upper-back mobility work to team training gave results similar to training alone²². Neither group came out ahead in strength, mobility or spike and serve speed²².
In a crossover trial (each person tries each option) of 15 active college-aged adults, one session raised extension and rightward turning more than brisk walking²³. The gains were within the minimal detectable change, the smallest change larger than measurement error²³.
A non-randomised study of 252 office workers had a waiting control group²⁴. After 12 weeks of whole-body device stretching, 14.7 percent fewer reported upper back pain²⁴.

One participant in the office study stopped because of upper-back discomfort²⁴.

What reviews found about posture

Reviews on the upper-back curve split stretching from strengthening^4,25:

In a review of healthy people, four stretching studies of the upper back and neck showed no effect on posture. Strengthening showed a very large effect there, across 10 studies⁴.
One review of exercise trials on spinal curves found a large effect of exercise on the kyphosis angle across seven trials²⁵. It suggests strengthening may matter more than stretching; the single trial in it that used stretching alone showed no gain²⁵.
In 28 trials in 901 people with upper crossed syndrome, corrective exercise programmes reduced the kyphosis angle, though results varied widely²⁶. Effects on function and pain were inconclusive²⁶.

Progressive overload

The doses studies used

Each study below reports the dose it used^1,22:

Study	Hold and repetitions	How often and how long
50 people, the named stretch	20 to 30 seconds, three times, 30 seconds rest¹	Three sets a session, five a week, four weeks¹
One man, over a roller	Five sets of 30 repetitions a day⁷	Ten days⁷
Office workers, extension exercises	Two sets of 15 repetitions of 10 seconds²	40 minutes, three times a week, six weeks²
Neck pain, kneeling rotation	Held 10 seconds, 10 times each side, two sets³	Three days a week, four weeks³
Volleyball players, eight exercises	15 repetitions each²²	Daily, four weeks²²
Office workers, device stretching	Five positions, each twice for 20 seconds²⁴	10 minutes, twice a week, 12 weeks²⁴

Where intensity changed

The volleyball programme used no progressive overload²². In the office-worker extension trial, the shoulder-blade exercises gained dumbbells after four weeks². In 11 older women with a mildly rounded upper back, a daily 20-minute foam roller routine grew harder over eight weeks²⁷. Their shoulder flexion range rose from about 149 to 155 degrees²⁷. The abstract describes no comparison group, and the curve itself was not among the measures retaken²⁷.

Connections

The muscles and joints it moves

The rhomboids draw the shoulder blade back toward the spine, the opposite of this stretch^1,8. So do the middle fibres of the trapezius⁹. The rib cage limits how far the thoracic part of the spine bends¹².

The posture review of healthy people is also on the doorway pec stretch page; it found stretching programmes did not change spinal posture⁴. In one study of the upper trapezius stretch, healthy adults in their 20s and 40s had lower systolic pressure after a session²⁸.